eWeek ran an article yesterday titled Linux Losing Market Share to Windows Server. The article quoted IDC sales figures. There is a real problem counting this way. Quoting from the article:
IDC analyst Al Gillen pointed out that the number of servers shipped does not perfectly equal the number of operating systems in the market. This is particularly the case with Linux where a substantial portion of the overall market opportunity comes from deployments aboard recycled servers, PCs and workstations deployed as servers, and Linux deployed as a guest operating system.
“This does not contradict any trending taking place on server hardware,” Gillen said.
He added: “But we do need to remember that the Linux software ecosystem does not track exactly the same as does x86 hardware shipments.”
I should note that eWeek often functions as a cheerleading section for Microsoft. Take, for example, the links to other articles with such catchy titles as:
- Windows Server Woos Linux Customers
- Is open source dying?
- Windows Server 2008 features address the Linux challenge.
In 2004 and 2005 I did five months of contract work for a large insurance company in Cincinnati. Their methodology for server purchasing is one I’ve seen time and time again in corporate America. They would underestimate their needs or else wait until the last minute to make purchases. They bought their servers based on what was in stock or what they could get a corporate discount on. They didn’t care if the servers were going to be used for Windows or Linux. They took them with whatever was on them, which was always Windows, and then promptly erased the hard drives and loaded a corporate standard image of whichever OS they wanted. Each and every server they loaded Linux on was counted as a Windows server when sales figures are tallied by organizations like IDC.
I find the claim that Windows now owns 70% of the server market hard to believe. Granted, I don’t work in Windows-only shops but I wonder how many of those really exist. Most large organizations have a mix of Windows, UNIX, and Linux. I also suspect the 20% figure IDC assigns to Linux is accurate in terms of server sales but grossly underreports what percentage of servers actually run Linux.
This leads me to a question I just don’t have a good answer for. How can Linux’ real market share and penetration be accurately measured?